January 10, 2014 | John Reynders has quite the pharma pedigree, having held roles at AstraZeneca, Johnson and Johnson, and Eli Lilly spanning informatics to companion diagnostics to neuroscience biomarkers. Before his turn in pharma, he was part of the team building informatics capability at Celera after the genome was assembled, and before that spent about eight years at Los Alamos National Laboratory.
Last fall, he traded his big pharma hat for the CIO position at Moderna, a privately held biotech based in Cambridge, Ma., focusing on messenger RNA therapeutics. Before the holidays, Reynders sat down with Bio-IT World Editor Allison Proffitt to talk about his vision for the three-year-old company. One that depends heavily on the cloud.
Bio-IT World: John, can you give me an overview of what you’re doing at Moderna?
John Reynders: Yeah, sure. So my role here at Moderna I’m Chief Information Officer, responsible for informatics, IT, and automation. I’ve been here about three months [as of November 2013]… One of the things I find really exciting about the opportunity here at Moderna is not only is it a transformative scientific platform—in terms of messenger RNA serving as a therapeutic and helping the body to make its own medicines—but the informatics that support actually designing these constructs. mRNA, in a sense, is a piece of software that’s telling the body, the hardware, how to program to make therapeutics and proteins that are its own medicine. There’s quite a bit of modeling, quite a bit of informatics, quite a bit of automation to bring that to light. So it’s an exciting challenge. Great to be part of it.
BITW: In March 2013, while you were still at AstraZeneca, the pharma announced a major agreement with Moderna starting with an upfront $240 million payment. Was your move a part of that deal?
Reynders: Oh, no. Basically, Stéphane [Bancel, Moderna’s President and Founding Chief Executive Officer] and I know each other from Eli Lilly and we sort of go back, and Stéphane had reached out to me well before the AstraZeneca deal was announced. As exciting as it was at AstraZeneca with all the technology opportunities there, what we were doing with big data, as I learned more about what Moderna was trying to do, it became very exciting, the opportunity. And actually, I had made my decision to join Moderna and it wasn’t until then that I learned about the deal and I thought, oh, okay, well, that’s really nice to know and sort of confirming evidence. But no, I was in discussions with Stéphane well before I knew about the agreement.
BITW: You’ve characterized your job at Moderna as, “creating a fully cloud-based biotech.” Can you tell me a little bit about what your goal is there? How does the cloud-based biotech fit in with RNA therapeutics?
Reynders: Sure. I mean one of the things that was really exciting about joining Moderna was the opportunity from kind of de novo—from scratch—to design informatics, a technology platform for a next-generation biotech. And one of the things that was very clear is there’s no way you’d be building up your own datacenters, building up your own internal on-premise applications. What you’re able to do in the cloud now is rather impressive. A lot of folks get distracted and think, oh, the cloud is all about saving money and doing IT cost-effectively. That’s certainly part of it. But the most valuable part of it is how fast you can move, the speed with which you can deploy new capabilities, the agility or you can change things and entire huge components of applications that are out there that can be readily and quickly assembled. So it’s challenging for an established firm or biotech because there’s a lot of legacy and moving to cloud, it’s something you can do around the edges. But for us, we made the commitment to doing everything in the cloud and other than some specific systems onsite to support labs, basically, we have all of our calculations, all of our production systems, our ERP [enterprise resource planning] and G&A systems, all served out of a variety of cloud solutions that are serving us very well.
BITW: But Moderna has a wet lab. You can’t do that in a cloud.
Reynders: Well, no, no. We’re designing and making messenger RNA therapeutics. So we certainly have laboratories. There is a strong component of lab systems, there’s a strong component of informatics that’s required for design, so all the in silico components whether it’s to design the mRNA at the beginning, whether it’s understanding how the mRNA construct is flowing through production, whether it’s tracking all of these pieces and parts as they’re coming together. That’s all done through the cloud, so everything to do with designing, tracking, and analytics. Other than a few systems that need to be directly bolted onto an instrument (to, for example, analyze certain imaging features), everything else pretty much we’ve taken up into the cloud. But indeed, no, we’re fully integrated in terms of being able to do everything from chemistries, to mRNA science, production, QC, screening, in vitro, in vivo, we have all those capabilities end-to-end in-house.
BITW: Who are you using for the cloud? Is this a private cloud, or are you using a commercial vendor?
Reynders: Oh, we’re using a constellation of partners. So I’ve been very impressed with Amazon. We set up a virtual private cloud with Amazon. They’ve done quite a bit to make sure their solutions are HIPAA compliant, recently, they have secured FedRAMP certification. We use Office 365 for all of our collaboration communications technology, whether it’s Lync or SharePoint. And then we’re also looking at a variety of bespoke, or specialized cloud solutions. We’re deploying SilkRoad Heartbeat as an HRIS (human resources information system), Greenlight for eLearning, and RedCarpet for onboarding—all cloud-based solutions. And we have also decided on our financial platform which will be NetSuite—again, a cloud based solution. We’re also architecting to ensure all these cloud solutions are able to integrate on top of one another. There’re some very strong platforms that are best-in-class. I wouldn’t propose to put everything on say Google or Force.com or Amazon, but I think each is able to serve a solution in the ecosystem very effectively and to complement each other very well. One of the things about the overall cloud architecture, is pulling all the pieces together.
BITW: You mentioned that for big pharma it would be hard for everything to go on the cloud because you can do it around the edges, but generally you’re so established, entrenched. What are the other differences you’re seeing as you come away from J&J and Lilly and AstraZeneca to this organization?
Reynders: I guess some of the differences are, you’re able to from the beginning think about how would you design a data architecture, an application architecture where the pieces can talk to one another. Something that I think has been challenging for large pharma is, either you have geographic separation, you have functional separation and whether it’s on-prem or off-prem, you always are trying to ask these integrated questions.
The harder problem is you have a lot of diversity of data in imaging data, genetic data, clinical data. The kinds of critical questions we’re asking of that data, the translational questions for example, you have to pull together lots of different kinds of data to ask these questions. And you have these perennial sort of silo-to-silo challenges that can exist in big pharma or big biotech.
Something we spent a lot of time on here at Moderna, is designing from the start. Here is what the entire data architecture needs to look like. Here’s the chemistry system, here’s the mRNA science and sequencing system, here’s production. We thought about ontologies, we’ve thought about the data architecture. When all of this is assembled, we can ask those integrated questions. So it’s very hard to integrate on the backend. You hear about all the silos and how it’s time to take a wrecking ball through them to make them talk to one another, so that’s one thing I’d say is very different. We’re in a position to architect from the get-go, here’s how all these important pieces of the puzzle need to talk to each other at once. And I think if you chat with any informatics or IT colleagues in big pharma, one of the top challenges you’ll typically find on the list is, how do we get our data to talk to each other?
Bio-IT World: What are some of the pitfalls you are protecting against as you build your system from the beginning?
Reynders: If you look at one of the pitfalls we have, it’s this big data myopia where you’re looking only at your class of data, but no matter how deep and sophisticated we become in understanding genotypes unless we also connect that to understanding genomics and pathways, it has limited value. Any silo by itself, no matter how deep, is not able to solve many of the complex heterogeneous data questions that we have. So it really starts by having hypothesis-driven questions that pull together the right data, rather than emerging questions where we go down one data silo or pull it together and hope something emerges.
Bio-IT World: What’s your end goal?
Reynders: At the end of the day, this is all about, “How do you enable effective decisions?” Of all the systems that we’re producing here, the front door is trying to understand, say, the key question in a nucleotide chemistry. Or a colleague is trying to understand the dynamics of a sequence. It’s all about solving very specific questions and making, frankly, what’s happening under the hood transparent. When [researchers] are asking those questions, here’s everything that’s happening to actually connect that piece of data to that piece of data so it never becomes a roadblock. So a scientist can follow as rapidly as possible their path and intuition to ask one question which leads to another question and leads to another question. You never want the informatics system or the IT system to say, “Oh, wait a second, now you need to go jump to this other environment and ask these other questions because of this other query.” It all starts with a question and a scientific path of inquiry and making sure that whether someone is trying to pull up a terabyte of data or you’re trying to combine all sorts of stuff that’s from different classes of data, they aren’t seeing that—they can just concentrate on their question.
That’s the beauty of, I would say, doing big data right. If you’re doing it right, then you never know what’s happening under the hood—and can just focus on pursuing your scientific insight.